Package  Description 

de.lmu.ifi.dbs.elki.algorithm.clustering.kmeans.initialization 
Initialization strategies for kmeans.

Modifier and Type  Class and Description 

class 
FarthestPointsInitialMeans<O>
KMeans initialization by repeatedly choosing the farthest point (by the
minimum distance to earlier points).

class 
FarthestSumPointsInitialMeans<O>
KMeans initialization by repeatedly choosing the farthest point (by the
sum of distances to previous objects).

class 
KMeansPlusPlusInitialMeans<O>
KMeans++ initialization for kmeans.

class 
OstrovskyInitialMeans<O>
Ostrovsky initial means, a variant of kmeans++ that is expected to give
slightly better results on average, but only works for kmeans and not for,
e.g., PAM (kmedoids).

class 
PredefinedInitialMeans
Run kmeans with prespecified initial means.

class 
RandomlyChosenInitialMeans<O>
Initialize Kmeans by randomly choosing k existing elements as initial
cluster centers.

class 
RandomNormalGeneratedInitialMeans
Initialize kmeans by generating random vectors (normal distributed
with \(N(\mu,\sigma)\) in each dimension).

class 
RandomUniformGeneratedInitialMeans
Initialize kmeans by generating random vectors (uniform, within the value
range of the data set).

class 
SampleKMeansInitialization<V extends NumberVector>
Initialize kmeans by running kmeans on a sample of the data set only.

Copyright © 2019 ELKI Development Team. License information.